Distributed Augmentation, Hypersweeps, and Branch Decomposition of Contour Trees for Scientific Exploration

Area of Science:

• Topological Data Analysis
• Scientific Visualization
• High-Performance Computing

Background:

• Contour trees are essential for analyzing scalar field topology but pose computational challenges for large datasets.
• Existing distributed contour tree methods require further computation for geometric properties and branch decomposition.
• Effective analysis and visualization of large-scale scientific data using contour trees remain a significant challenge.

Purpose of the Study:

• To develop distributed algorithms for augmenting and decomposing distributed contour trees.
• To enable parallel computation of geometric properties for enhanced scientific exploration.
• To support efficient contour extraction and analysis in large-scale scientific visualization.

Main Methods:

• Introduced distributed algorithms for augmentation, hypersweeps, and branch decomposition.
• Enabled parallel computation of geometric properties on distributed contour trees.
• Evaluated parallel performance and applied algorithms to identify and extract important contours.

Main Results:

• Demonstrated the effectiveness of distributed algorithms for computing geometric properties.
• Showcased the capability to use distributed contour trees as query structures for scientific exploration.
• Successfully identified and extracted important contours for scientific visualization using the developed methods.

Conclusions:

• The introduced distributed algorithms significantly advance the scalability of contour tree analysis.
• These methods facilitate more effective exploration and visualization of large-scale scientific data.
• The work provides a robust framework for parallel computation and analysis of distributed contour trees.